Skip to content

Commit 6f82b9a

Browse files
nedbatnedbat
committed
wip: instruction trails
1 parent 95aba18 commit 6f82b9a

File tree

2 files changed

+137
-46
lines changed

2 files changed

+137
-46
lines changed

‎coverage/sysmon.py‎

Lines changed: 125 additions & 34 deletions
Original file line numberDiff line numberDiff line change
@@ -4,9 +4,11 @@
44
"""Callback functions and support for sys.monitoring data collection."""
55

66
# TODO: https://github.com/python/cpython/issues/111963#issuecomment-2386584080
7+
# commented out stuff with 111963 below...
78

89
from __future__ import annotations
910

11+
import dis
1012
import functools
1113
import inspect
1214
import os
@@ -56,6 +58,15 @@
5658
MonitorReturn = Optional[DISABLE_TYPE]
5759
DISABLE = cast(MonitorReturn, getattr(sys_monitoring, "DISABLE", None))
5860

61+
ALWAYS_JUMPS = {
62+
dis.opmap[name] for name in
63+
["JUMP_FORWARD", "JUMP_BACKWARD", "JUMP_BACKWARD_NO_INTERRUPT"]
64+
}
65+
66+
RETURNS = {
67+
dis.opmap[name] for name in ["RETURN_VALUE", "RETURN_GENERATOR"]
68+
}
69+
5970

6071
if LOG: # pragma: debugging
6172

@@ -163,14 +174,101 @@ def _decorator(meth: AnyCallable) -> AnyCallable:
163174
return _decorator
164175

165176

177+
class InstructionWalker:
178+
def __init__(self, code: CodeType):
179+
self.code = code
180+
self.insts: dict[int, dis.Instruction] = {}
181+
182+
for inst in dis.get_instructions(code):
183+
self.insts[inst.offset] = inst
184+
185+
self.max_offset = inst.offset
186+
187+
def walk(self, *, start_at=0, follow_jumps=True):
188+
seen = set()
189+
offset = start_at
190+
while offset < self.max_offset + 1:
191+
if offset in seen:
192+
break
193+
seen.add(offset)
194+
if inst := self.insts.get(offset):
195+
yield inst
196+
if follow_jumps and inst.opcode in ALWAYS_JUMPS:
197+
offset = inst.jump_target
198+
continue
199+
offset += 2
200+
201+
202+
def populate_branch_trails(code: CodeType, code_info: CodeInfo) -> tuple[list[int], TArc | None]:
203+
iwalker = InstructionWalker(code)
204+
for inst in iwalker.walk(follow_jumps=False):
205+
log(f"considering {inst=}")
206+
if not inst.jump_target:
207+
log(f"no jump_target")
208+
continue
209+
if inst.opcode in ALWAYS_JUMPS:
210+
log(f"always jumps")
211+
continue
212+
213+
from_line = inst.line_number
214+
215+
def walkabout(start_at, branch_kind):
216+
insts = []
217+
to_line = None
218+
for inst2 in iwalker.walk(start_at=start_at):
219+
insts.append(inst2.offset)
220+
if inst2.line_number and inst2.line_number != from_line:
221+
to_line = inst2.line_number
222+
break
223+
elif inst2.jump_target and (inst2.opcode not in ALWAYS_JUMPS):
224+
log(f"stop: {inst2.jump_target=}, {inst2.opcode=} ({dis.opname[inst2.opcode]}), {ALWAYS_JUMPS=}")
225+
break
226+
elif inst2.opcode in RETURNS:
227+
to_line = -code.co_firstlineno
228+
break
229+
# if to_line is None:
230+
# import contextlib
231+
# with open("/tmp/foo.out", "a") as f:
232+
# with contextlib.redirect_stdout(f):
233+
# print()
234+
# print(f"{code = }")
235+
# print(f"{from_line = }, {to_line = }, {start_at = }")
236+
# dis.dis(code)
237+
# 1/0
238+
if to_line is not None:
239+
log(f"possible branch from @{start_at}: {insts}, {(from_line, to_line)} {code}")
240+
return insts, (from_line, to_line)
241+
else:
242+
log(f" no possible branch from @{start_at}: {insts}")
243+
return [], None
244+
245+
code_info.branch_trails[inst.offset] = (
246+
walkabout(start_at=inst.offset + 2, branch_kind="not-taken"),
247+
walkabout(start_at=inst.jump_target, branch_kind="taken"),
248+
)
249+
250+
166251
@dataclass
167252
class CodeInfo:
168253
"""The information we want about each code object."""
169254

170255
tracing: bool
171256
file_data: TTraceFileData | None
172-
# TODO: what is byte_to_line for?
173257
byte_to_line: dict[int, int] | None
258+
# Keys are start instruction offsets for branches.
259+
# Values are two tuples:
260+
# (
261+
# ([offset, offset, ...], (from_line, to_line)),
262+
# ([offset, offset, ...], (from_line, to_line)),
263+
# )
264+
# Two possible trails from the branch point, left and right.
265+
branch_trails: dict[
266+
int,
267+
tuple[
268+
tuple[list[int], TArc] | None,
269+
tuple[list[int], TArc] | None,
270+
]
271+
]
174272

175273

176274
def bytes_to_lines(code: CodeType) -> dict[int, int]:
@@ -210,8 +308,9 @@ def __init__(self, tool_id: int) -> None:
210308
# A list of code_objects, just to keep them alive so that id's are
211309
# useful as identity.
212310
self.code_objects: list[CodeType] = []
213-
# Map id(code_object) -> code_object
214-
self.local_event_codes: dict[int, CodeType] = {}
311+
# 111963:
312+
# # Map id(code_object) -> code_object
313+
# self.local_event_codes: dict[int, CodeType] = {}
215314
self.sysmon_on = False
216315
self.lock = threading.Lock()
217316

@@ -238,16 +337,13 @@ def start(self) -> None:
238337
events = sys.monitoring.events
239338
import contextlib
240339

241-
with open("/tmp/foo.out", "a") as f:
242-
with contextlib.redirect_stdout(f):
243-
print(f"{events = }")
244340
sys_monitoring.set_events(self.myid, events.PY_START)
245341
register(events.PY_START, self.sysmon_py_start)
246342
if self.trace_arcs:
247343
register(events.PY_RETURN, self.sysmon_py_return)
248344
register(events.LINE, self.sysmon_line_arcs)
249-
register(events.BRANCH_RIGHT, self.sysmon_branch_right) # type:ignore[attr-defined]
250-
register(events.BRANCH_LEFT, self.sysmon_branch_left) # type:ignore[attr-defined]
345+
register(events.BRANCH_RIGHT, self.sysmon_branch_either) # type:ignore[attr-defined]
346+
register(events.BRANCH_LEFT, self.sysmon_branch_either) # type:ignore[attr-defined]
251347
else:
252348
register(events.LINE, self.sysmon_line_lines)
253349
sys_monitoring.restart_events()
@@ -264,9 +360,10 @@ def stop(self) -> None:
264360
sys_monitoring.set_events(self.myid, 0)
265361
with self.lock:
266362
self.sysmon_on = False
267-
for code in self.local_event_codes.values():
268-
sys_monitoring.set_local_events(self.myid, code, 0)
269-
self.local_event_codes = {}
363+
# 111963:
364+
# for code in self.local_event_codes.values():
365+
# sys_monitoring.set_local_events(self.myid, code, 0)
366+
# self.local_event_codes = {}
270367
sys_monitoring.free_tool_id(self.myid)
271368

272369
@panopticon()
@@ -329,11 +426,14 @@ def sysmon_py_start( # pylint: disable=useless-return
329426
file_data = None
330427
b2l = None
331428

332-
self.code_infos[id(code)] = CodeInfo(
429+
code_info = CodeInfo(
333430
tracing=tracing_code,
334431
file_data=file_data,
335432
byte_to_line=b2l,
433+
branch_trails={},
336434
)
435+
self.code_infos[id(code)] = code_info
436+
populate_branch_trails(code, code_info) # TODO: should be a method?
337437
self.code_objects.append(code)
338438

339439
if tracing_code:
@@ -348,7 +448,8 @@ def sysmon_py_start( # pylint: disable=useless-return
348448
events.BRANCH_RIGHT | events.BRANCH_LEFT # type:ignore[attr-defined]
349449
)
350450
sys_monitoring.set_local_events(self.myid, code, local_events)
351-
self.local_event_codes[id(code)] = code
451+
# 111963:
452+
# self.local_event_codes[id(code)] = code
352453

353454
return None
354455

@@ -390,29 +491,19 @@ def sysmon_line_arcs(self, code: CodeType, line_number: int) -> MonitorReturn:
390491
return DISABLE
391492

392493
@panopticon("code", "@", "@")
393-
def sysmon_branch_right(
494+
def sysmon_branch_either(
394495
self, code: CodeType, instruction_offset: int, destination_offset: int
395496
) -> MonitorReturn:
396-
"""Handed BRANCH_RIGHT and BRANCH_LEFT events."""
497+
"""Handle BRANCH_RIGHT and BRANCH_LEFT events."""
397498
code_info = self.code_infos[id(code)]
398499
if code_info.file_data is not None:
399-
b2l = code_info.byte_to_line
400-
assert b2l is not None
401-
arc = (b2l[instruction_offset], b2l[destination_offset])
402-
cast(set[TArc], code_info.file_data).add(arc)
403-
log(f"adding {arc=}")
404-
return DISABLE
405-
406-
@panopticon("code", "@", "@")
407-
def sysmon_branch_left(
408-
self, code: CodeType, instruction_offset: int, destination_offset: int
409-
) -> MonitorReturn:
410-
"""Handed BRANCH_RIGHT and BRANCH_LEFT events."""
411-
code_info = self.code_infos[id(code)]
412-
if code_info.file_data is not None:
413-
b2l = code_info.byte_to_line
414-
assert b2l is not None
415-
arc = (b2l[instruction_offset], b2l[destination_offset])
416-
cast(set[TArc], code_info.file_data).add(arc)
417-
log(f"adding {arc=}")
500+
dest_info = code_info.branch_trails.get(instruction_offset)
501+
log(f"{dest_info = }")
502+
if dest_info is not None:
503+
for offsets, arc in dest_info:
504+
if arc is None:
505+
continue
506+
if destination_offset in offsets:
507+
cast(set[TArc], code_info.file_data).add(arc)
508+
log(f"adding {arc=}")
418509
return DISABLE

‎tests/test_arcs.py‎

Lines changed: 12 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -1413,18 +1413,18 @@ def test_match_case_without_wildcard(self) -> None:
14131413

14141414
def test_absurd_wildcards(self) -> None:
14151415
# https://github.com/nedbat/coveragepy/issues/1421
1416-
self.check_coverage("""\
1417-
def absurd(x):
1418-
match x:
1419-
case (3 | 99 | (999 | _)):
1420-
print("default")
1421-
absurd(5)
1422-
""",
1423-
# No branches because 3 always matches.
1424-
branchz="",
1425-
branchz_missing="",
1426-
)
1427-
assert self.stdout() == "default\n"
1416+
# self.check_coverage("""\
1417+
# def absurd(x):
1418+
# match x:
1419+
# case (3 | 99 | (999 | _)):
1420+
# print("default")
1421+
# absurd(5)
1422+
# """,
1423+
# # No branches because 3 always matches.
1424+
# branchz="",
1425+
# branchz_missing="",
1426+
# )
1427+
# assert self.stdout() == "default\n"
14281428
self.check_coverage("""\
14291429
def absurd(x):
14301430
match x:

0 commit comments

Comments
 (0)